Retrievals of vertically resolved aerosol microphysical particle parameters with regularization from spaceborne Aerosol and Carbon dioxide Detection Lidar (ACDL)

Zhixian Bi, Jianbo Hu, Yuan Xie, Decang Bi, Xiaopeng Zhu, Jiqiao Liu, Weibiao Chen2025 · DOI

TL;DR

This paper introduces a novel regularization method to derive aerosol microphysical properties (like effective radius and concentration) directly from spaceborne multi-wavelength lidar data. By incorporating a third wavelength (1572 nm), the method significantly improves the accuracy of retrieving coarse-mode aerosols, demonstrating ACDL's potential for global aerosol characterization.

What Problem Does This Paper Address?

Accurately understanding aerosols is crucial for climate modeling, as they significantly impact Earth's radiation budget. However, aerosols are highly variable and complex, making global monitoring challenging. While spaceborne lidars offer a promising solution for continuous, global observations, deriving detailed microphysical properties (like particle size distribution) from their data has been an ongoing challenge. Existing methods often rely on assumptions or limited wavelengths, leading to uncertainties, especially for larger aerosol particles. This paper aims to overcome these limitations by developing a more robust retrieval algorithm for spaceborne lidar.

Methods

The core of this work is an improved regularization algorithm designed to invert lidar optical data (backscatter and extinction coefficients) into aerosol microphysical parameters. Unlike previous approaches that might use fewer wavelengths or rely on specific assumptions, this method leverages the three wavelengths (532 nm, 1064 nm, and 1572 nm) provided by the Aerosol and Carbon dioxide Detection Lidar (ACDL). The algorithm approximates the aerosol volume concentration distribution using a linear superposition of base functions and then applies regularization to solve the ill-posed inverse problem. A key innovation is the inclusion of the 1572 nm channel, which is shown to significantly enhance the retrieval of coarse-mode aerosols. The process also involves creating a lookup table (LUT) and averaging multiple inversion results to improve accuracy and avoid underfitting.

Datasets

  • ACDL (Aerosol and Carbon dioxide Detection Lidar) data: This is the primary dataset used for retrieving aerosol microphysical properties. The paper utilizes simulated and actual observational data from ACDL.
  • CALIPSO (Cloud Aerosol Lidar and Infrared Pathfinder Satellite Observation) data: Used for comparison with ACDL inversion results.
  • LIVAS (Lidar climatology of Vertical Aerosol Structures) database: Also used for comparison, providing mean aerosol optical properties.
  • AERONET (Aerosol Robotic Network) data: Used to generate simulated aerosol models (urban, dust, marine, biomass burning) for testing the algorithm.

Key Insights and Ideas

  • The power of three wavelengths: Adding the 1572 nm channel to the traditional 532 nm and 1064 nm significantly boosts the retrieval accuracy for larger, coarse-mode aerosol particles. This is a crucial finding for understanding aerosols that heavily influence climate.
  • Regularization as a robust tool: The regularization approach proves effective in handling the inherent ill-posed nature of inverting lidar optical data into microphysical parameters, providing stable results even with complex aerosol mixtures.
  • ACDL's potential for global characterization: The study demonstrates that ACDL, as the first spaceborne high-spectral-resolution lidar, has the capability to provide vertically resolved aerosol microphysical information on a global scale.
  • Sensitivity to input data: While the algorithm is robust, systematic errors in the input optical data (especially at 532 nm extinction) can lead to significant uncertainties in the retrieved microphysical parameters, highlighting the importance of accurate lidar measurements.

Key Results

  • Simulation accuracy: For monomodal aerosols, retrieval errors are within 15% for effective radius ($r_{eff}$), 30% for surface area concentration ($S_t$), and 35% for volume concentration ($V_t$). For bimodal aerosols, these errors increase to 18-35% for $r_{eff}$, 35% for $S_t$, and up to 60% for $V_t$.
  • Sensitivity analysis: Systematic errors of ±20% in input optical data lead to parameter uncertainties below 60%.
  • Case study profiles: Typical retrieved effective radii include urban (~0.47 µm), smoke (~0.57 µm), dust (~0.61 µm), and marine (~0.83 µm).
  • Comparison with CALIPSO/LIVAS: ACDL inversion results show high consistency with CALIPSO and LIVAS for marine and dust aerosols. Urban and smoke retrievals from ACDL tend to be slightly larger.

Limitations and Future Work

  • Bimodal distribution challenges: Inversions for bimodal lognormal distributions show some oscillations at the edges of the retrieval interval, attributed to insufficient optical data input.
  • Fine mode information loss: While the algorithm excels at coarse mode retrieval, some information on fine mode particles is inevitably lost.
  • Idealized simulations: The sensitivity analysis was based on idealized simulations. Future work will focus on extending these retrievals to real observational data.
  • Future research directions: The authors suggest constructing scattering lookup tables for different aerosol shapes based on depolarization ratio data and improving the inversion quality for fine mode particles. They also aim to further enhance the accuracy of ACDL data products for aerosol particle size parameters.

Summary updated

Abstract

Using an improved regularization method, we attempt to derive microphysical parameters (effective radius 𝑟𝑒𝑓𝑓, surface area concentration 𝑆𝑡, volume concentration 𝑉𝑡) of aerosol particle size distribution directly from the detection results of Aerosol and Carbon dioxide Detection Lidar (ACDL), which is the first spaceborne high spectral resolution lidar. The backscatter and extinction coefficients at 532 nm, 1064 nm, 1572 nm are adopted for regularization algorithm. Preliminary simulations for different aerosol types demonstrate the algorithm performance of the 3α+3β optical data combination. For monomodal aerosols, the retrieval errors are constrained within 15 % for 𝑟𝑒𝑓𝑓, 30 % for 𝑆𝑡, and 35 % for 𝑉𝑡. In bimodal cases, errors increase to 18–35 % for 𝑟𝑒𝑓𝑓, 35 % for 𝑆𝑡, and up to 60 % for 𝑉𝑡. Sensitivity analysis confirms that systematic errors of ±20 % in input optical data induce parameter uncertainties below 60 %. Case studies reveal four typical aerosols profiles: urban (𝑟𝑒𝑓𝑓~0.5 μm), smoke (𝑟𝑒𝑓𝑓~0.6 μm), dust (𝑟𝑒𝑓𝑓~0.65 μm), and marine (𝑟𝑒𝑓𝑓~0.85 μm). The inversion 𝑟𝑒𝑓𝑓 is compared with CALIPSO and LIVAS, which confirms high consistency for marine and dust, while urban and smoke retrievals show slightly larger. The inclusion of 1572 nm significantly enhances coarse-mode retrieval accuracy. The error statistics of the simulations and the actual comparison results show that the proposed inversion algorithm can reliably derive the particle size distribution parameters from the spaceborne multi-wavelength lidar ACDL. This work provides preliminary validation of ACDL's capability to retrieve vertically resolved global aerosol microphysical characterization.
BibTeX
@misc{bi2025retrievals,
  author = {Zhixian Bi and Jianbo Hu and Yuan Xie and Decang Bi and Xiaopeng Zhu and Jiqiao Liu and Weibiao Chen},
  title = {{Retrievals of vertically resolved aerosol microphysical particle parameters with regularization from spaceborne Aerosol and Carbon dioxide Detection Lidar (ACDL)}},
  year = {2025},
  doi = {10.5194/egusphere-2025-4208},
}
Figure